5 th SASTech 211, Khavaran Higher-education Institute, Mashhad, Iran. May 12-14. 1 Sensorless Torue and Speed Control of Traction Permanent Magnet Synchronous Motor for Railway Applications based on Model Reference Adaptive System Mohammad Ali Akbari 1, Vahid Irani 2, Mohammad Ali Sandidzadeh 3 Iran University of Science and Technology m_akbari85@rail.iust.ac.ir, iranidtc@yahoo.com, khaburi@iust.ac.ir Paper Reference Number: 62-981 Name of the Presenter: Mohammad Ali Akbari Baseri Abstract In recent years, by improving performance of magnet material and decreasing the cost, interest is growing in utilization of Traction Permanent Magnet Synchronous Motor (Traction PMSM) in propulsion systems. In this paper a speed sensorless control method for Traction PMSM, based on Model Reference Adaptive System (MRAS) is represented. In the proposed method the stator current is used to estimate the motor speed. The direct axis reference current (id) is considered to be zero for vector control. By implementing this assumption on stator current, only uadrature axis current (i) is used for estimation the speed. Therefore the error would be simplified and the error controller, which is a PI controller, follows the rotor speed easily and causes the error to be zero. Stability analysis for proposed MRAS would be inspected and proved by Popov Hyperstability theory. Simulation results clearly show the suitable performance of the estimator in low and high speeds. Key words: Traction PMSM, Vector Control, Speed Estimation, MRAS 1. Introduction Currently, DC and induction motors are used as traction motors for railway vehicles, but in recent years PMSMs have received significant attention for use in railway due to their uniue characteristics including lower energy consumption, higher efficiency, lower weight-to-power ratio, reliable operation, low noise production, robustness, etc. With the emergence progress in product of new permanent magnet materials, modern Traction PMSMs has been developed with high power and torue. Today, giant manufacturers of train are doing vast investigations on using PMSM in their products. In 27 Toshiba has developed the world s first PMSM propulsion system in commercial service in Tokyo subway. Vector control method needs the information about rotor position and speed. Insufficient space to install the sensor in some cases and also reduction in system costs raised the attention to sensorless speed control for PMSMs. Several methods have been presented for speed and position estimation, such as Back EMF position estimators [1], high freuency
5 th SASTech 211, Khavaran Higher-education Institute, Mashhad, Iran. May 12-14. 2 signal injection [2], Advanced Kalman Filter [3], advanced Leonberger Observer, Sliding Mode Observer [4], etc. Each of methods have their own advantages and weak points. The methods based on MRAS, depending on the state variable considered for system, would have different estimation performances. Reference [5] uses reactive power to estimate the speed. In reference [6], the stator current of has been used to estimate the speed, but the error is complicated and PI controller is not able to estimate the speed easily. In this paper, a speed estimation method based on MRAS has been presented which uses the stator current as a state variable for speed estimation. In the field oriented control, the direct axis reference current is considered to be zero. This assumption on MRAS simplifies the error and eventually causes the error controller to be able to estimate the speed easily and dynamically as well. The stability for proposed system has been examined using Popov Hyperstability theory. 2.Mathematic Model of PMSM Voltage euations of PMSM in the synchronous rotating reference frame are: ud Rid pd u Ri p d (1) Where Li d d d af Li (2) Where u d, u, i d, i are the d- axis stator voltages and currents, respectively; R is stator resistance; p is differential coefficient; λ d, λ are the d- axis stator magnetic flux, respectively; ω is electrical motor speed; L d, L are the d- axis stator inductances, respectively; λ af is coupling flux linkage of rotor on stator. The electromagnetic torue is given by: Where P is the number of pole pairs. 3 Te Paf i (3) 2 3. Speed Estimation Methods based on MRAS Model Reference Adaptive System theory is based on calculating an arbitrary state variable using two models. One of them is reference model, and another is comparative model. The unknown parameter would be estimated using the error between two models. Also the reference model has to be independent of estimated variable and in contrary; the comparative model should be dependent to the estimated variable. The error signal is given to the compare mechanism unit. The estimated variable (which is speed in this case)
5 th SASTech 211, Khavaran Higher-education Institute, Mashhad, Iran. May 12-14. 3 would be applied to the comparative model by compare unit output. This loop continues until the compare unit output finally euals the reference model output, and estimated variable reaches its final value. Stability of this system is examined using Popov Hyperstability criterion as well. This method is easy and needs simple computing. Fig. 1 shows the overall structure of MRAS. Fig. 1. Block Diagram of MRAS Selecting the state variable as the output for both reference and comparative model enables us to develop different structures of MRAS. In some cases, the adaptive reference model is configured using stator flux, in other cases the stator currents has been used as state variables. Other structures using the reactive power as a state variable. For each state variable used in MRAS, a uniue statement term of error criterion would be obtained which is applied to the compare mechanism in order to estimate the considered variable. 4. Estimator Stability Examination If the stator current vector is considered as state variable, then according to machine euations, the stator current would be calculated as follows: di dt 1 ( u Ri Lsid af ) (4) L s Estimated current is given by: di dt 1 ( u R i Lsid af ) (5) L s Finally, the error of estimated current can be achieved as: d( i i ) 1 R( i i ) ( )( Ldid af ) dt L s (6) In the field oriented control, since the direct current is zero, the error would be obtained as follows: d( i i) 1 R( i i ) ( ) af dt L s (7)
5 th SASTech 211, Khavaran Higher-education Institute, Mashhad, Iran. May 12-14. 4 With the following definition for error, the state euation of error would be obtained conseuently: ( i i) (8) ' A w (9) The criterion for the error value to gradually approach to zero, is based on the fact that error system state euations should be stable; in other word, all of the poles in state euations should be on the left side. The second criterion says the nonlinear feedback should hold true in the following ineuality: t T 2 (, t) v wdt (1) In which, 2 is a positive real number. Assume that: ( DI) v D v (11) Finally the following term would be obtained after substitution of above terms in the Popov criterion. t af (, t) ( i )( ) i dt (12) Ls Also considering PI controller for speed estimation, we would have: t F v t d F v t 1(,, ) 2(, ) () (13) Now, with insertion of above estimated speed in the Popov criterion, and solving the terms for Popov ineuality, the following euation would be obtained for speed estimation: (14) t ( ) ( af ) af k () i i i dt k p i i Ls Ls And finally the estimated speed can be written as: t ( ) ( k ) () i i i dt k p i i (15) 5. Simulation Results Traction PMSM Vector Control without speed sensor using by MRAS method has been simulated using Matlab/Simulink. Results confirm the validity of proposed method. Table 1 shows the Traction PMSM parameters used in the simulation. Rating of Traction PMSM Stator resistance d- axis inductance 16 kw, 9 V, 75 RPM.5 Ω L d = L =2. 5 mh
5 th SASTech 211, Khavaran Higher-education Institute, Mashhad, Iran. May 12-14. 5 Total moment of inertia 2.2 kg.m 2 Friction factor.5558 N.m.s Number of pole pairs 8 PM flux λ af =.3 V.s Nominal torue T N = 1 N.m Table 1: specification of proposed Traction PMSM The motor speed and load torue diagrams vs time are illustrated in fig. 2. The simulation time set to 6 s. Speed (rpm) & Torue (N.m) 1 5-5 Reference Speed Load Torue -1 1 2 3 4 5 6 Fig. 2. Motor reference speed and load torue diagrams vs time Train survey the distance between two stations in 3 general regions; the first one is positive acceleration (speed increasing), and the second one is constant speed, and the last one is negative acceleration (braking). In the beginning of this survey the load torue on axle of traction motor has its maximum value. With increasing the speed, load torue is decreased and finally in the braking region the load torue become negative. Reference, real and estimated speeds of traction motor are illustrated in fig. 3. 8 Speed (rpm) 6 4 2-2 Reference Estimated Real 1 2 3 4 5 6 Fig. 3. Reference, real and estimated speeds of traction motor
5 th SASTech 211, Khavaran Higher-education Institute, Mashhad, Iran. May 12-14. 6 Real speed follows the reference speed very well with proper adjustment of PI controllers coefficients. Furthermore as it clear in fig. 3 the estimated speed follows the real speed acceptable. The electromagnetic torue of traction motor and d- axis currents are indicated in figs. 4 and 5, respectively. Torue (N.m) 1-1 Electromagnetic Torue Load Torue 1 2 3 4 5 6 Fig. 4. Load and electromagnetic torues 5 I (A) -5 1 2 3 4 5 6 5 Id (A) -5-1 1 2 3 4 5 6 Fig. 5. Direct and uadrature axis currents As we expected the direct axis current almost zero and the uadrature axis current follows electromagnetic torue with constant ratio. Fig. 6 shows the system s speed step response, the reference step speed of 75 rpm has been applied in.1s to drive; also the torue eual to 5 N.m is applied to the motor. The estimated speed, reference speed and real speed of motor are shown in Fig. 6. As it can be seen, the estimated speed follows the real speed very well. Also the d- currents are shown in fig. 7.
5 th SASTech 211, Khavaran Higher-education Institute, Mashhad, Iran. May 12-14. 7 1 Speed (rpm) I ( A ) 5 Reference Estimated Real -5.2.4.6.8 1 4 2 Fig. 6. Speed Step Response alongside the Estimated Speed -2.2.4.6.8 1 5 I d ( A ) -5.2.4.6.8 1 Fig. 7. Direct and uadrature currents Motor performance at zero speed and speed estimation error has been shown in Fig. 8. 6 Speed (rpm) 4 2 Reference Estimated Real -2.5 1 1.5 2 2.5 3
5 th SASTech 211, Khavaran Higher-education Institute, Mashhad, Iran. May 12-14. 8 Speed Estimation Error (rpm).5 -.5.5 1 1.5 2 2.5 3 Fig. 8. Operation of Speed Estimator and Estimation Error in Zero Speed 5. Conclusion A sensorless vector control of Traction PMSM, which uses stator current as state variable for speed estimation, is presented in this paper. In proposed method the direct axis reference current (id) is considered to be zero and only uadrature axis current (i) is used for estimation the speed. This assumption lead to error simplification and caused the speed estimator have a suitable performance. In proposed method the estimated speed follows the real speed very well, furthermore the estimator perform well at low speeds. References [1] B.N. Mobarakeh, F.M. Tabar, F. M. Sargos, (24). Back-EMF estimation based sensorless control of PMSM: robustness with respect to measurement errors and inverter irregularities. in: Conf. Rec. of 39th IAS Annual Meeting, vol. 3, pp. 1858 1865. [2] B.K. Bose. (1996) Power Electronics and Variable Freuency Drives Technology and Applications. IEEE press, New York. [3] S. Bologani, R. Oboe, M. Zigliotto. (1999). Sensorless full digital PMSM drive with EKF estimation of speed and rotor position. IEEE Trans. Ind. Electron. 46 (1),pp. 184 191. [4] Z.M.A. Peixo, F.M. Freitas Sa, P.F. Seixas, B.R. Menezes, P.C. Cortizo, W.S. Lacerda. (1995). Application of sliding mode observer for induced e.m.f., position and speed estimation of permanent magnet motors. in: Proceedings of 1995 International Conference on Power Electronics and Drive Systems, vol. 2, pp. 599 64. [5] Suman Maiti, Chandan Chakraborty. (29). Simulation studies on model reference adaptive controller based speed estimation techniue for vector controlled PMSM drive. ELSEVIER, simulation modeling practice and theory 17, 585-596. [6] Young Sam Kim, Sang Kyoon Kim, Young Ahn Kwon. (23). Mras based sensorless control of permanent magnet synchronous motor. SICE 23 Annual Conference, vol. 2, pp. 1632-1637.